Bug#41756 "Strange error messages about locks from InnoDB".
In JT_EQ_REF (join_read_key()) access method,
don't try to unlock rows in the handler, unless certain that
a) they were locked
b) they are not used.
Unlocking of rows is done by the logic of the nested join loop,
and is unaware of the possible caching that the access method may
have. This could lead to double unlocking, when a row
was unlocked first after reading into the cache, and then
when taken from cache, as well as to unlocking of rows which
were actually used (but taken from cache).
Delegate part of the unlocking logic to the access method,
and in JT_EQ_REF count how many times a record was actually
used in the join. Unlock it only if it's usage count is 0.
Implemented review comments.
mysql-test/r/innodb_lock_wait_timeout_1.result:
Update results (Bug41756).
mysql-test/t/innodb_lock_wait_timeout_1.test:
Add a test case (Bug#41756).
sql/item_subselect.cc:
Complete struct READ_RECORD initialization with a new
member to unlock records.
sql/records.cc:
Extend READ_RECORD API with a method to unlock read records.
sql/sql_select.cc:
In JT_EQ_REF (join_read_key()) access method,
don't try to unlock rows in the handler, unless certain that
a) they were locked
b) they are not used.
sql/sql_select.h:
Add members to TABLE_REF to count TABLE_REF buffer usage count.
sql/structs.h:
Update declarations.
Bug#41756 "Strange error messages about locks from InnoDB".
In JT_EQ_REF (join_read_key()) access method,
don't try to unlock rows in the handler, unless certain that
a) they were locked
b) they are not used.
Unlocking of rows is done by the logic of the nested join loop,
and is unaware of the possible caching that the access method may
have. This could lead to double unlocking, when a row
was unlocked first after reading into the cache, and then
when taken from cache, as well as to unlocking of rows which
were actually used (but taken from cache).
Delegate part of the unlocking logic to the access method,
and in JT_EQ_REF count how many times a record was actually
used in the join. Unlock it only if it's usage count is 0.
Implemented review comments.
mysql-test/r/bug41756.result:
Add result file (Bug#41756)
mysql-test/t/bug41756-master.opt:
Use --innodb-locks-unsafe-for-binlog, as in 5.0 just
using read_committed isolation is not sufficient to
reproduce the bug.
mysql-test/t/bug41756.test:
Add a test file (Bug#41756)
sql/item_subselect.cc:
Complete struct READ_RECORD initialization with a new
member to unlock records.
sql/records.cc:
Extend READ_RECORD API with a method to unlock read records.
sql/sql_select.cc:
In JT_EQ_REF (join_read_key()) access method,
don't try to unlock rows in the handler, unless certain that
a) they were locked
b) they are not used.
sql/sql_select.h:
Add members to TABLE_REF to count TABLE_REF buffer usage count.
sql/structs.h:
Update declarations.
values return too many records
WHERE clauses with "outer_value_list NOT IN subselect" were
handled incorrectly if the outer value list contained multiple
items where at least one of these could be NULL. The first
outer record with NULL value was handled correctly, but if a
second record with NULL value existed, the optimizer would
choose to reuse the result it got on the last execution of the
subselect. This is incorrect if the outer value list has
multiple items.
The fix is to make Item_in_optimizer::val_int (in
item_cmpfunc.cc) reuse the result of the latest execution
for NULL values only if all values in the outer_value_list
are NULL.
mysql-test/r/subselect3.result:
Added test for BUG#48177
mysql-test/t/subselect3.test:
Added test for BUG#48177
sql/item_cmpfunc.cc:
Make Item_in_optimizer::val_int (in item_cmpfunc.cc) reuse the result of the latest execution for NULL values only if all values in the outer_value_list are NULL.
Detailed revision comments:
r6126 | vasil | 2009-10-30 10:36:07 +0200 (Fri, 30 Oct 2009) | 45 lines
branches/zip: Merge r6112:6125 from branches/5.1:
(skipping r6122 and r6123, Jimmy says these are already present and need
not be merged):
------------------------------------------------------------------------
r6122 | jyang | 2009-10-30 05:18:38 +0200 (Fri, 30 Oct 2009) | 7 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
M /branches/5.1/mysql-test/innodb_bug44369.result
M /branches/5.1/mysql-test/innodb_bug44369.test
M /branches/5.1/mysql-test/innodb_bug46000.result
M /branches/5.1/mysql-test/innodb_bug46000.test
branches/5.1: Chnage WARN_LEVEL_ERROR to WARN_LEVEL_WARN
for push_warning_printf() call in innodb.
Fix Bug#47233: Innodb calls push_warning(MYSQL_ERROR::WARN_LEVEL_ERROR)
rb://170 approved by Marko.
------------------------------------------------------------------------
r6123 | jyang | 2009-10-30 05:43:06 +0200 (Fri, 30 Oct 2009) | 8 lines
Changed paths:
M /branches/5.1/os/os0proc.c
branches/5.1: In os_mem_alloc_large(), if we fail to attach
the shared memory, reset memory pointer ptr to NULL, and
allocate memory from conventional pool. This is a port
from branches/zip.
Bug #48237 Error handling in os_mem_alloc_large appears to be incorrect
rb://198 Approved by: Marko
------------------------------------------------------------------------
r6125 | vasil | 2009-10-30 10:31:23 +0200 (Fri, 30 Oct 2009) | 4 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
branches/5.1:
White-space fixup.
------------------------------------------------------------------------
Detailed revision comments:
r6121 | sunny | 2009-10-30 01:42:11 +0200 (Fri, 30 Oct 2009) | 7 lines
branches/zip: This test has been problematic for sometime now. The underlying
bug is that the data dictionaries get out of sync. In the AUTOINC code we
try and apply salve to the symptoms. In the past MySQL made some unrelated
change and the dictionaries stopped getting out of sync and this test started
to fail. Now, it seems they have reverted that changed and the test is
passing again. I suspect this is not he last time that this test will change.
r6124 | jyang | 2009-10-30 10:02:31 +0200 (Fri, 30 Oct 2009) | 5 lines
branches/zip: Correct the bug number for -r6109 change
from # 48273 to #48237
Detailed revision comments:
r6114 | vasil | 2009-10-29 18:43:51 +0200 (Thu, 29 Oct 2009) | 6 lines
branches/zip:
* Add ChangeLog entries for latest changes
* Obey alphabetical order in the list of the files
* White-space fixup
Detailed revision comments:
r6113 | vasil | 2009-10-29 18:15:50 +0200 (Thu, 29 Oct 2009) | 93 lines
branches/zip: Merge r5912:6112 from branches/5.1:
(after this merge the innodb-autoinc test starts to fail, but
I commit anyway because it would be easier to investigate the
failure this way)
------------------------------------------------------------------------
r5952 | calvin | 2009-09-22 19:45:07 +0300 (Tue, 22 Sep 2009) | 7 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
branches/5.1: fix bug#42383: Can't create table 'test.bug39438'
For embedded server, MySQL may pass in full path, which is
currently disallowed. It is needed to relax the condition by
accepting full paths in the embedded case.
Approved by: Heikki (on IM)
------------------------------------------------------------------------
r6032 | vasil | 2009-10-01 15:55:49 +0300 (Thu, 01 Oct 2009) | 8 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
branches/5.1:
Fix Bug#38996 Race condition in ANALYZE TABLE
by serializing ANALYZE TABLE inside InnoDB.
Approved by: Heikki (rb://175)
------------------------------------------------------------------------
r6045 | jyang | 2009-10-08 02:27:08 +0300 (Thu, 08 Oct 2009) | 7 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
A /branches/5.1/mysql-test/innodb_bug47777.result
A /branches/5.1/mysql-test/innodb_bug47777.test
branches/5.1: Fix bug #47777. Treat the Geometry data same as
Binary BLOB in ha_innobase::store_key_val_for_row(), since the
Geometry data is stored as Binary BLOB in Innodb.
Review: rb://180 approved by Marko Makela.
------------------------------------------------------------------------
r6051 | sunny | 2009-10-12 07:05:00 +0300 (Mon, 12 Oct 2009) | 6 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
M /branches/5.1/mysql-test/innodb-autoinc.result
M /branches/5.1/mysql-test/innodb-autoinc.test
branches/5.1: Ignore negative values supplied by the user when calculating the
next value to store in dict_table_t. Setting autoincrement columns top negative
values is undefined behavior and this change should bring the behavior of
InnoDB closer to what users expect. Added several tests to check.
rb://162
------------------------------------------------------------------------
r6052 | sunny | 2009-10-12 07:09:56 +0300 (Mon, 12 Oct 2009) | 4 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
M /branches/5.1/mysql-test/innodb-autoinc.result
M /branches/5.1/mysql-test/innodb-autoinc.test
branches/5.1: Reset the statement level autoinc counter on ROLLBACK. Fix
the test results too.
rb://164
------------------------------------------------------------------------
r6053 | sunny | 2009-10-12 07:37:49 +0300 (Mon, 12 Oct 2009) | 6 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
M /branches/5.1/mysql-test/innodb-autoinc.result
M /branches/5.1/mysql-test/innodb-autoinc.test
branches/5.1: Copy the maximum AUTOINC value from the old table to the new
table when MySQL does a CREATE INDEX ON T. This is required because MySQL
does a table copy, rename and drops the old table.
Fix Bug#47125: auto_increment start value is ignored if an index is created and engine=innodb
rb://168
------------------------------------------------------------------------
r6076 | vasil | 2009-10-14 19:30:12 +0300 (Wed, 14 Oct 2009) | 4 lines
Changed paths:
M /branches/5.1/row/row0mysql.c
branches/5.1:
Fix typo.
------------------------------------------------------------------------
Detailed revision comments:
r6110 | marko | 2009-10-29 12:44:57 +0200 (Thu, 29 Oct 2009) | 2 lines
branches/zip: Makefile.am (INCLUDES): Merge a change from MySQL:
Use $(srcdir)/include instead of $(top_srcdir)/storage/innobase/include.
r6111 | marko | 2009-10-29 13:04:11 +0200 (Thu, 29 Oct 2009) | 33 lines
branches/zip: Fix corruption of buf_pool->LRU_old and improve debug assertions.
This was reported as Issue #381.
buf_page_set_old(): Assert that blocks may only be set old if
buf_pool->LRU_old is initialized and buf_pool->LRU_old_len is nonzero.
Assert that buf_pool->LRU_old points to the block at the old/new boundary.
buf_LRU_old_adjust_len(): Invoke buf_page_set_old() after adjusting
buf_pool->LRU_old and buf_pool->LRU_old_len, in order not to violate
the added assertions.
buf_LRU_old_init(): Replace buf_page_set_old() with a direct
assignment to bpage->old, because these loops that initialize all the
blocks would temporarily violate the assertions about
buf_pool->LRU_old.
buf_LRU_remove_block(): When setting buf_pool->LRU_old = NULL, also
clear all bpage->old flags and set buf_pool->LRU_old_len = 0.
buf_LRU_add_block_to_end_low(), buf_LRU_add_block_low(): Move the
buf_page_set_old() call later in order not to violate the debug
assertions. If buf_pool->LRU_old is NULL, set old=FALSE.
buf_LRU_free_block(): Replace the UNIV_LRU_DEBUG assertion with a
dummy buf_page_set_old() call that performs more thorough checks.
buf_LRU_validate(): Do not tolerate garbage in buf_pool->LRU_old_len
even if buf_pool->LRU_old is NULL. Check that bpage->old is monotonic.
buf_relocate(): Make the UNIV_LRU_DEBUG checks stricter.
buf0buf.h: Revise the documentation of buf_page_t::old and
buf_pool_t::LRU_old_len.
Detailed revision comments:
r6109 | jyang | 2009-10-29 10:37:32 +0200 (Thu, 29 Oct 2009) | 7 lines
branches/zip: In os_mem_alloc_large(), if we fail to attach
the shared memory, reset memory pointer ptr to NULL, and
allocate memory from conventional pool.
Bug #48237 Error handling in os_mem_alloc_large appears to be incorrect
rb://198 Approved by: Marko
Detailed revision comments:
r6108 | calvin | 2009-10-29 07:58:04 +0200 (Thu, 29 Oct 2009) | 5 lines
branches/zip: close file handle when building with UNIV_HOTBACKUP
The change does not affect regular InnoDB engine. Confirmed by
Marko.
Detailed revision comments:
r6107 | marko | 2009-10-28 16:10:34 +0200 (Wed, 28 Oct 2009) | 5 lines
branches/zip: buf_page_set_old(): Improve UNIV_LRU_DEBUG diagnostics
in order to catch the buf_pool->LRU_old corruption reported in Issue #381.
buf_LRU_old_init(): Set the property from the tail towards the front
of the buf_pool->LRU list, in order not to trip the debug check.
Detailed revision comments:
r6103 | marko | 2009-10-26 15:46:18 +0200 (Mon, 26 Oct 2009) | 4 lines
branches/zip: row_ins_alloc_sys_fields(): Zero out the system columns
DB_TRX_ID, DB_ROLL_PTR and DB_ROW_ID, in order to avoid harmless
Valgrind warnings about uninitialized data. (The warnings were
harmless, because the fields would be initialized at a later stage.)
Detailed revision comments:
r6101 | jyang | 2009-10-23 11:45:50 +0300 (Fri, 23 Oct 2009) | 7 lines
branches/zip: Update test result with the WARN_LEVEL_ERROR
to WARN_LEVEL_WARN change. This is the same result as
submitted in rb://172 review, which approved by Sunny Bains
and Marko.
Detailed revision comments:
r6100 | jyang | 2009-10-22 06:51:07 +0300 (Thu, 22 Oct 2009) | 6 lines
branches/zip: As a request from mysql, WARN_LEVEL_ERROR cannot
be used for push_warning_* call any more. Switch to
WARN_LEVEL_WARN. Bug #47233.
rb://172 approved by Sunny Bains and Marko.
Detailed revision comments:
r6099 | jyang | 2009-10-22 05:58:39 +0300 (Thu, 22 Oct 2009) | 7 lines
branches/zip: Port bug #46000 related changes from 5.1 to zip
branch. Due to different code path for creating index in zip
branch comparing to 5.1), the index reserved name check function
is extended to be used in ha_innobase::add_index().
rb://190 Approved by: Marko
Detailed revision comments:
r6095 | vasil | 2009-10-19 16:04:59 +0300 (Mon, 19 Oct 2009) | 7 lines
branches/zip:
Fix Bug#47808 innodb_information_schema.test fails when run under valgrind
by using the wait_until_rows_count macro that loops until the number of
rows becomes 14 instead of sleep 0.1, which is obviously very fragile.
Detailed revision comments:
r6064 | calvin | 2009-10-13 18:23:35 +0300 (Tue, 13 Oct 2009) | 4 lines
branches/zip: non-functional changes
Changes from MySQL to fix build issue.
r6065 | inaam | 2009-10-13 20:43:13 +0300 (Tue, 13 Oct 2009) | 7 lines
branches/zip rb://182
Call fsync() on datafiles after a batch of pages is written to disk
even when skip_innodb_doublewrite is set.
Approved by: Heikki
Detailed revision comments:
r6049 | vasil | 2009-10-09 19:05:26 +0300 (Fri, 09 Oct 2009) | 7 lines
branches/zip:
Fix compilation warning in Hot Backup:
innodb/fil/fil0fil.c: In function 'fil_load_single_table_tablespace':
innodb/fil/fil0fil.c:3253: warning: format '%lld' expects type 'long long int', but argument 6 has type 'ib_int64_t'
Detailed revision comments:
r6048 | vasil | 2009-10-09 08:42:55 +0300 (Fri, 09 Oct 2009) | 16 lines
branches/zip:
When scanning a directory readdir() is called and stat() after it,
if a file is deleted between the two calls stat will fail and the
whole precedure will fail. Change this behavior to continue with the
next entry if stat() fails because of nonexistent file. This is
transparent change as it will make it look as if the file was deleted
before the readdir() call.
This change is needed in order to fix
https://svn.innodb.com/mantis/view.php?id=174
in which we need to abort if os_file_readdir_next_file()
encounters "real" errors.
Approved by: Marko, Pekka (rb://177)
Detailed revision comments:
r6046 | pekka | 2009-10-08 12:24:56 +0300 (Thu, 08 Oct 2009) | 3 lines
branches/zip: Revert r6044 which added os_file_is_same() function
(issue#186). This functionality is moved to Hot Backup source tree.
Detailed revision comments:
r6044 | pekka | 2009-10-06 17:44:54 +0300 (Tue, 06 Oct 2009) | 5 lines
branches/zip:
Add os_file_is_same() function for Hot Backup (inside ifdef UNIV_HOTBACKUP).
This is part of the fix for Issue #186.
Note! The Windows implementation is incomplete.
When a sessione is closed, all temporary tables of the session are automatically
dropped and are binlogged. But it will be binlogged with wrong database names when
the length of the temporary tables' database names are greater than the
length of the current database name or the current database is not set.
Query_log_event's db_len is forgot to set when Query_log_event's db is set.
This patch wrote code to set db_len immediately after db has set.
- disabled main.innodb_bug47777.test with InnoDB plugin
until fix for plugin is applied.
- disabled main.innodb-autoinc.test (failing)
- re-enabled main.innodb_bug39438.test
- added error message suppression to innodb_bug39438, as
requested by InnoDB/Oracle
- reverted change to main.innodb_bug34300 as plugin specific.
Detailed revision comments:
r6127 | vasil | 2009-10-30 11:18:25 +0200 (Fri, 30 Oct 2009) | 18 lines
branches/5.1:
Backport c6121 from branches/zip:
------------------------------------------------------------------------
r6121 | sunny | 2009-10-30 01:42:11 +0200 (Fri, 30 Oct 2009) | 7 lines
Changed paths:
M /branches/zip/mysql-test/innodb-autoinc.result
branches/zip: This test has been problematic for sometime now. The underlying
bug is that the data dictionaries get out of sync. In the AUTOINC code we
try and apply salve to the symptoms. In the past MySQL made some unrelated
change and the dictionaries stopped getting out of sync and this test started
to fail. Now, it seems they have reverted that changed and the test is
passing again. I suspect this is not he last time that this test will change.
------------------------------------------------------------------------
Detailed revision comments:
r6123 | jyang | 2009-10-30 05:43:06 +0200 (Fri, 30 Oct 2009) | 8 lines
branches/5.1: In os_mem_alloc_large(), if we fail to attach
the shared memory, reset memory pointer ptr to NULL, and
allocate memory from conventional pool. This is a port
from branches/zip.
Bug #48237 Error handling in os_mem_alloc_large appears to be incorrect
rb://198 Approved by: Marko
Detailed revision comments:
r6052 | sunny | 2009-10-12 07:09:56 +0300 (Mon, 12 Oct 2009) | 4 lines
branches/5.1: Reset the statement level autoinc counter on ROLLBACK. Fix
the test results too.
rb://164
r6053 | sunny | 2009-10-12 07:37:49 +0300 (Mon, 12 Oct 2009) | 6 lines
branches/5.1: Copy the maximum AUTOINC value from the old table to the new
table when MySQL does a CREATE INDEX ON T. This is required because MySQL
does a table copy, rename and drops the old table.
Fix Bug#47125: auto_increment start value is ignored if an index is created and engine=innodb
rb://168
Detailed revision comments:
r6051 | sunny | 2009-10-12 07:05:00 +0300 (Mon, 12 Oct 2009) | 6 lines
branches/5.1: Ignore negative values supplied by the user when calculating the
next value to store in dict_table_t. Setting autoincrement columns top negative
values is undefined behavior and this change should bring the behavior of
InnoDB closer to what users expect. Added several tests to check.
rb://162
